`The Unique Structure of the Present Great Andamanese: An Overview of the Grammar

Abstract Great Andamanese is an isolate and belongs to the sixth language family of India (Abbi 2009). Present-day Great Andamanese (PGA), a koinèized version of the North Great Andamanese languages is a head-marking polysynthetic and agglutinative language with an SOV pattern, and has a very elaborate system for marking inalienability (Abbi 2006, 2010) nested in seven possessive markers designating different body-divisions. These markers are further grammaticalized in the language and appear as proclitics which classify a large number of lexical items as dependent categories. The author proposes that the Great Andamanese conceptualize their world through these interdependencies and hence the grammar of the language encodes this important phenomenon in every grammatical category expressing referential, attributive and predicative meaning. These are very unusual features never reported earlier in grammars of languages of the world and thus, indicate very old structures in the chain of language evolution.

1. Introduction

The Andaman Islands are comprised of a cluster of approximately 550 islands, rocks and rocky outcrop running from north to south and located southeast of the Indian sub- continent in the Bay of Bengal. They are separated from the Malay Peninsula by the

Andaman Sea , an extension of the Bay of Bengal, and are part of the Union Territory of the Andaman and Nicobar Islands belonging to India (Map 1). Geographically, the

Andaman Islands are closer to Myanmar and Indonesia than to mainland India. However, contact between the Andamanese and the populations of the neighboring countries has not been established till date. The capital city of the Andaman Islands is Port Blair,

1 situated in the south of the Islands at a distance of 1255 km from Kolkata and 1190 km from Chennai.

There are ten languages in the Great Andamanese family, which can be grouped into three varieties: southern, central and northern. These are: Aka-Bea, Aka-Bale, the

2 southern variety; Aka-Pucikwar [known as Pujjukar in the current spoken language], Aka-Kol, Aka-Kede, Aka-Jowoi, as the central variety; and Aka-Jeru, Aka-Bo, Aka-Kora [known as Khora by the present speakers] and Aka-Cari [known as Sare by the present speakers] a northern variety. Except for Jeru and Sare 1 all Great Andamanese languages are now extinct. Not all languages were mutually intelligible with each other as the languages of the Great Andamanese tribes formed a ‘‘linguistic continuum”, so that each language was closely related to its neighbor on each side but those at the extreme ends of the geographic continuum were mutually unintelligible. Hence, Aka-Cari (Map 2), a North Great Andamanese language, was mutually unintelligible with Aka Bea, the southern variety. The present-day Great Andamanese 2 language is a mixture of four northern varieties 3 with sporadic interferences from the central variety such as Aka Pucikwar. Linguists consider Great Andamanese as a language isolate because no links to any other language family have been established so far.

1 There is one speaker who hails from the background of Sare, however speaks the present form of the Great Andamanese language .

2 I will use the term Present-day Great Andamanese (PGA) for the present form of the language and avoid referring to it by any of the four languages from which it draws its resources so as not to empower one language over the other. It is spoken in the Strait Island and some parts of Port Blair.

3 The last few generations of Great Andamanese speakers are descendents of intermarriages among North Andamanese tribes. The Government of India encouraged this practice in order to preserve their dwindling numbers when the entire population was settled on ‘Strait Island’.

3

Map 2 Courtesy: www.andaman.org

Though the PGA is characterized by a mixture of the linguistic features of four to five varieties of Great Andamanese languages, what we notice in today’s Great Andamanese speech is a kind of levelling of different linguistic systems. Perhaps several grammatical inputs have contributed to generate the present language. The linguistic system of present Great Andamanese appears to be close to koine ization (Manoharan

4 1989). As the language is critically endangered, with just eight terminal speakers 4, it is very difficult to say how far it is mixed and what elements are mixed. However, the belief of Siegel (1985:363) that koineization results in the reduction and simplification of grammar is attested to by some areas in the grammar of Great Andamanese; though the verb morphology and the possession constructions are rather complex and elaborate.

The research reported here is based on the first-hand field data elicited during the period of 2001-2002 and 2005-2009. Several visits were made to the Strait Island and

Port Blair where the speakers of the language reside 5.

The latest research by Abbi in 2003 and 2006 show that Great Andamanese constitutes the sixth language family of India. Linguisti c researc h on th e survivin g language s of th e Andama n Islands reveal s littl e commonalit y betwee n Grea t Andamanes e and the languages of the Jaraw a-Ong e group. (Abb i, 2003 , 2006 , 2009). The Jarawa-

Onge group has been associated with Austronesian language family (Blevins 2007). Out of the ten varieties that once existed in the Great Andamanese family, we found traces of only four languages, i.e. Sare, Khora, Bo and Jero in today’s speech. The recent deaths of

4 This was the number of fluent speakers when we began our research in the island. There are only five speakers left now. Fortunately, we could interview some of the fluent speakers of the language when they were still alive. Special mention must be made of Jirake, the chief of the Great Andamanese tribe and Nao Jr. his younger brother, and Boa Sr who came from the Bo tribe. More than 50% of the current population of the Great Andamanese tribe consists of children below 14 years of age (See Abbi et al 2007) 5 The research was conducted in two phases. The first one was a pilot survey of the languages of the Andaman Islands supported and funded by the Max Planck Institute of Evolutionary Anthropology, Leipzig, Germany. The second phase–research was conducted at Jawaharlal Nehru University, New Delhi under the project Vanishing Voices of the Great Andamanese (VOGA) supported by the Hans Rausing Endangered Language Fund, SOAS, University of London.

5 the last speakers of Khora and Bo have left only Jero and Sare speakers who are not fully conversant with their respective languages but remember isolated words from their native tongues. The demographic scale of these islanders is inversely related to their degree of contact with mainlanders: the longer the contact, the smaller the population.

2 The sound system

We discuss in short the sound system of the PGA language in the following pages. Because of the fact that PGA is a moribund language with a few speakers left in the community and the fact that our speakers hail from diverse background, care has been taken to observe and report phonetic variation across speakers. This information will, perhaps, give an idea of phonemic grid of the extinct and dying languages of the Great Andamanese family.

2.1 Vowels PGA has a seven ‐vowel system, as shown in Table 1, and offers a large possibility of combinations in the area of vowel sequences or clusters, as represented in Table 2. We noted a high variation in the inventory of vowels and consonants among the speakers because of ‘koiné’ and the ‘mixed’ nature of the language. Another factor leading to such variation could be the fact that the language is on the verge of extinction and community members do not remember many words and their exact pronunciation, and hence, offer varied sounds for the same word. The indifference of the speakers towards the language could also lead to such variation. Despite such variation, phonemic inventories of vowels and consonants could be arrived at by eliciting minimal pairs for most of the sounds. Where minimal pairs were not available, the judgements of the native speakers about the phonological contrasts were taken into account.

6 Front Central Back Close i u Half close e o Half open Open

Table 1 Vowels of Great Andamanese

In the first phase of fieldwork (2001–2002), we recorded the mean mid central vowel [ ] in the data. However, subsequent visits to the speech community and the digital recording of the vowel indicated that the language lacks this central vowel. There is only one unrounded back vowel and that is a. The rendering of Hindi words such as bnao ‘make’, was also realized as banao . There are more archiphonemes in this language than any other language that I have come across. Hence, contrast is neutralized across speakers. For example, we noticed that though e and ε stand in contrast, speakers use them interchangeably in some words. Similar was the situation with back vowels o and , as well as with o and u, which were in free variation, at times within the speech of the same speaker, despite the fact that the two sounds do offer contrast in minimal pairs. We have tried to capture this variation wherever possible in the dictionary, with the source specified.

We could not attest length at the phonemic level. We have specified length where we thought it was important for the pronunciation of the word, especially in the combination of two vowels (see Table 2). Phonetically it has been specified by a colon mark [:] placed after the long vowel.

2.2 Vowel sequences/clusters

The language is rich in vowel sequences or vowel clusters. We could attest such clusters, involving both short and long varieties of vowels. Also, some speakers use an epenthetic semivowel between short vowels within a word. For example, some speakers use the palatal semivowel y as in ia > iya , or the rounded back semivowel w as in ua > uwa

7 perhaps because of the influence of Hindi. However, only a few speakers use such epenthetic insertions and they too do not use them consistently. One of our contact persons, Peje never used a semivowel between two vowels.

Front Back ia, iu, io , i:o, ie, ua, uo, u i:e, ei, eo, e , oa, o:a, o:ѐ, ou, oi, oe, o:e, oε εo, εo ѐi, :e, e ao, a:o, a:u, au, a ε, a:e, ae, ai

Table 2 Vowel clusters of Great Andamanese

Only one example of a three-vowel cluster εkεie ‘pick up in lap’ was noticed. As mentioned earlier, the phonemic status of the length of vowels is doubtful, as it varies from speaker to speaker; oco ‘net’ while o:c ‘net’. The long and short u varies freely before a final vowel in a situation of vowel cluster. Thus, there are variations among speakers in the renderings for ‘my ear’ r-bu:o and r-buo . We have tried our best to document all the variations so that the reader gets an idea of the variety of sounds available in present Great Andamanese. For details refer to A Dictionary of the Great Andamanese language by Anvita Abbi (in press).

2.3 Consonants

As far as consonants are concerned, we attested unique inventories of sounds, such as the evidence of bilabial fricatives, both voiced and voiceless β, φ, and labialized lateral lw (at least in one speaker, Peje). PGA does not attest the voiced velar sound g and the voiceless glottal fricative h. The former sound was available in languages such as Bo and Pujjukar, which are extinct now, and in the languages of south and middle Andaman. We could attest this sound in the speech and songs of Boa Sr., the only Bo speaker that we had. This and the fact that the name of one of our consultants was Golat means that the sound must have existed in one of the Great Andamanese languages spoken in the past.

8 Conversely, we noticed the acquisition of h from Hindi in the speech of some young speakers. The following sound sets are in free variation at the intra ‐community level, i.e. within the same clan.

[φ ~p ~ f]

w [ ~ l ~ w~ l ] [k ~ x] [s ~ ~ c ~ c] [r ~ ] [t ~ ]

Table 3 Phonetic variations across speakers

The sounds in free variation vary from speaker to speaker, thus confirming the hypothesis that the existing speakers of the language are not descendants of the speakers of one language but of speakers of different varieties of the same language family. In other words, PGA represents inter ‐group free variation. However, the substitution of sounds in free variation did not disturb the comprehension of the word in question. The intra ‐community variation renders a large number of sound inventories, as shown in Table 4. The non-phonemic sounds that occur with low frequency are given in brackets in the table. Our consultant, Peje, for example, does not have the lateral l in his verbal w repertoire as he invariably uses l in all positions where other members use l. This could be an idiosyncratic feature of Peje or could be relics of the extinct languages of the family.

Labio- Bilabial Dental Alveolar Retroflex Palatal Velar Dental p b t d c j k Plosive p t k

m n Nasal

(ײ) Trill r

9 () ( β) (f) s (x) Fricative

Lateral (l w) l Approximant (w) y Table 4 Consonants of Great Andamanese

Considering the speech of the elderly persons in the community one cannot fail to observe that due to contact with Hindi, speakers are losing the voiceless bilabial fricative

φ to voiceless bilabial aspirated obstruent p and voiced counterpart β to voiced bilabial obstruent b. One can safely say that the bilabial fricatives are on their way to extinction. Similarly, in the last 40 years, the voiceless velar fricative x has changed to the voiceless velar aspirated k. Only one speaker, a 76-year-old woman called Boro, who was

.and this has also been documented ,ײ originally a Khora speaker, used the retroflex trill

Boro died in November 2009. One of our main consultants, Lico, who calls herself a Sare speaker but whose language has traces of Pujjukar language because of her upbringing, used s wherever other speakers used . Our oldest speaker, Boa Sr., who was a Bo speaker, used c and c instead of the sibilant . We wish there were more speakers and the language was vibrant enough to give us a fairly represented speech profile. However, one can give the following pattern of variation for these sounds, tracing them back to their sources. The names of the source languages are given in parentheses. c/c (Bo) ~ s (Sare) ~ (Jero)

At times, there is inter-changeability of sounds within the speech of the same speaker. For instance, our main consultant, Nao Jr, often substituted the dental [t] for the retroflex [ ] and vice versa, as in the word ara ile mo ‘bladder’ for ara iletmo . All these variations exist despite the fact that the pair of sounds under consideration stand in contrast. Such variation in the speech of the same speaker can only be ascribed to the status of the language, which is dying rapidly, and is not used by speakers in their daily life. The disuse of the language seems to be the most significant reason for the variation in sounds, lexicon, as well as syntactic constructions.

10 3. Syllable structure

Syllables have the following structure:

(C) (V) (C) (C) V (C) (V)

Only the vowel (V) is obligatory in Great Andamanese. Furthermore, syllables can involve consonants (C) in the onset (i.e. beginning) or the coda (i.e. end) of the syllable. Vowel sequences or vowel clusters are very common and involve two syllabic peaks within a word. Consonant clusters are rare in the language. They occur either word initially or in-between a word between two vowels. They also occur across word boundary in compound words. No word ever ends in a consonant cluster. Most simple syllable structure, i.e. constituting of one obligatory vowel is seen in words denoting genitive affixes or clitics. As a consequence, the following syllable structures are possible in simple words: Forms Examples Meaning V -a ‘genitive affix’ VV ie ‘catch’, ‘give’ CV o ‘stay, ‘live’ CVV boi ‘ask’ CVVC uoc ‘hear’ CVC br ‘breeze’, ‘air’ VCV uno ‘sit down’ VVCV eole ‘see’ CVCV cone ‘go’ CVCVC ema ‘run’ CCVCV trpo ‘climb (tree)’ [rare] CVCCV jermo ‘worm’ Table 5 Possible syllable structures

As affixation and compounding are very productive word-formation devices in the language, a complex word can be as long as of five syllables: akakertotlco ‘uvula’ VCV -CVC -CVC -CVCVC 1- 2 3 4 5

11 bilikututo 'cobweb' CVCVCV -CVC -CV 1 2 3 4 5

Names of birds, insects, reptiles and other jungle creatures provide most of the complex words of long syllable structures constituting, at times, two consonant clusters within the same word, as in trkimo ‘long-tail cricket’ or kulemo ‘wasp ( Vespa affinis)’.

4. Typology of the language and word formation processes Present Great Andamanese is a head-marking polysynthetic and agglutinative language with two types of nouns: dependent and non-dependent. Most of the nouns that refer to the typical inalienably possessed items as well as those which refer to the objects or results of an action are marked for dependency. All body-part terms, kinship terms, part- to-whole, part-to-component, as well as nouns referring to time, direction, and depth are dependent nouns. The dependent nouns are further divided into seven classes, each defined by a distinct body part inalienability marker appearing as a proclitic. There are three numbers encoded in pronominal forms. However, nouns in general are not marked for duality and plurality. Number is marked for plurality on a few common nouns designating animacy features, e.g. ‘dogs’, ‘children’ etc.

A large number of morphemes, affixes, phonological words, clitics and incorporation can constitute a single phonological word . This word generally is a verb phrase. Great Andamanese is a prototypical “head marking” language where the verb complex includes a large amount of information in multi-morphemic strings that include subject and object pronominal prefixes or clitics, incorporated nominals in causative constructions, reflexive and reciprocal prefixes, as well as suffixes expressing tense, aspect and mood. Overt external NPs are present in addition to the verb complex .

12 However these are optional and often dropped in discourse. PGA is a verb final language. However, while the genitive phrase precedes the head noun (as is typical of verb-final languages), other modifiers follow the modified. The language is of the

S(ubject)-O(bject)-V(erb) type. Some unexpected features, such as the fact that morphological causative markers are always prefixed rather than suffixed to the verb raise questions about the typology of the original structure of the language and changes in the language during the course of its history.

PGA offers basically two different sets of subject markers, one for prototypical agent nominals -e suffixed to agent nouns and another, -bi , which is attached to the subjects of unergative intransitive verbs, the subjects of unaccusative verbs, and object nominals. It was observed that in discourse and in fast speech speakers tend to drop these markings. We symbolize this relationship of noun with the corresponding verbs as:

So = Sa = O = -bi or unmarked.

A = -e or unmarked.

Present Great Andamanese is thus, an incompletely ergative-absolutive language with a fluid S-marking.

4.1 Word Formation processes

Attributive adjectives and transitive verbs are generally preceded by a proclitic. Thus, we have kata ‘piece’ and ot ‐kata ‘dwarf’; and ot-lѐko ‘nude ’ and ot-lѐko-ke ‘to bare all’. These clitics are very important as any change in the clitic changes the meaning of the verb in question. For instance, εr-pk-e ‘take out from fire or any hot medium’ (imperative)

εt-pok-e ‘kill’ (imperative) Sometimes a new word is derived by adding negative marker or an adjective to an existing word. Thus, we have

13 εr‐ulu ‐phu o‐kara ‐cae CLT ‐eye ‐NEG CLT ‐nails ‐rotten blind maimed

Similarly, verbs can be derived from nouns by adding verbal elements, such as object clitic (in case of a transitive verb), mood, and tense markers.

ѐl ‘flower’ but it-ѐl-e

OBJ CLT –flower-imperative ‘Pluck it.’ o 'house' but obom

house-verb class-NON PAST ‘He lives.’ As PGA is an agglutinative language, words are constituted of several morphemes. Compounding is a very productive word formation process employed by this language. Surprisingly, reduplication of any type does not exist in the language. We could attest only one word lele ‘swing’. Thus, Present Great Andamanese defies one of the norms of language universals.

5. Possessive constructions Possessive constructions are one of the most intriguing and interesting aspects of the language. The language offers a wide variety of possessives unlike any other Indian language. The distinctions between various forms do not depend only upon simple binary oppositions of physical alienability/inalienability but on various diverse ethno-semantic categories defining the relation between the possessor and possessed nouns. Factors such as part-whole relationship, intimate/non-intimate relations, human/non-human relations, inextricably linked relations, permanency of the relationship between the possessor and the possessed, the notion of non transferability of the possessed entity from the possessor, and finally, the conceptual dependency between the possessor and possessed play an important role in deciding the appropriate possessive marker which is termed as

‘inalienability marker’ (INA for short) in the grammar. This is prefixed to the possessed

14 nominal. Morpho-syntactically, the marker serves the function of a PROCLITIC and relates the possessor and the possessed. The INA marker is appropriately selected by the semantic categorization of the two nouns that it relates to.

5.1 Primary Possession

One major deciding criterion that determines the choice of a particular INA is the partonomy of the body. Under various possessive forms lies the semantic typology for the categorization of the human body parts. PGA maintains seven divisions (Table 5) within the partonomy of body and then further extends the INA markings for these seven divisions to a variety of other terms including: kin terms, spatial relational terms, closely related object terms, human attribute/propensity terms, and terms concerning actions, manner and states (denoted by verbs). However, their biggest function is classificatory . In this function they are lexically determined by the possessed noun and result in distinct and overt markings of possession symbolized by INA . The INA markers are grammaticalized further in the language and are prefixed as proclitics to transitive and intransitive verb roots, or those verbs that necessarily require a patient noun, viz: ‘sing’ ‘fight’ etc. Some of these markers are also used in describing kinship terms. See Table 6.

5.1.1 Affixation

As the language is dependent marked, the INA is prefixed to the possessed pronominal root/noun.

5.1.1 Schema of possessives in Present-day Great Andamanese

(S1) Possessor Pronominal clitic/Noun + INA + Possessed noun The structure can be abbreviated as:

(S2) [PP + INA + Poss N] Thus: (1) h= 6a=mimi

1SG -INA -mother

6 The symbol = signifies that the previous sound or phoneme is a clitic.

15 ‘My mother.’

(2) h= εr=co

1SG -INA -head

‘My head.’

(3) des -ot=bec

Des-INA -hair

‘Hair of Des’.

5.2 Possessor noun : The fundamental division of animacy plays an important role in deciding the phonetic shape of the base form of the INA . If the possessor noun is non- animate the marker is prefixed by a dental consonant t-, while with all animate possessors, both human and non-human, the INA begins with a vowel. Thus, the INA marker ara- and ot-,which are indicators of animate possessors, are rendered as tara - and tot- respectively if the possessors are inanimate beings. Hence, (4) ra-εr=co

pig-INA -head ‘Pig’s head.’ but (5) ra-tεr=co

pig-INA -head ‘Pig’s head.’ (When it has been cut.) (6) moco-ara=mo o

hen-INA -leg ‘Hen’s leg.’ but (7) moco-tara=mo o

hen-INA -leg ‘Chicken’s leg.’ (after it has been cut) (8) kri-ot=car

dugong-INA -chest ‘Dugong’s chest.’ but

16 (9) kri-tot=car

dugong-INA -chest ‘Chest of dugong.’ (Cut, or shown from a distance.) (10) fc-ta=po

vessel-INA -mouth ‘Mouth of a vessel.’ (11) bun-tr=pir

shell-INA -edge 'Sharp edge of a shell.’

5.3 Possessed noun

No body part term can be rendered without an appropriate INA marker. The unique feature of the language is that each division/area of the body parts is symbolized differently by a distinct INA marker. This implies that the Great Andamanese visualize their world from a vantage point of view centered in the ‘self’ or ‘ego’. But, each division within the realm of inalienability is designated on a hierarchical scale. The same division is carried over in perceiving human relations and other objects which are conceptually dependent. The binary distinction of alienable versus inalienable does not exist in a strict sense as used in English. There are only a few prototypical alienable nouns in PGA which exhibit external possession while all others fall within the purview of inalienability.

17 Classes Partonomy of human body INA MARKERS

1 mouth and its semantic extension a- 2 major external body parts ϯr-

3 extreme ends of the body like toes and fingernails o-

4 bodily products and part-whole relationship ut-

5 organs inside the body e-

6 parts designating round shape/sexual organs ara-

7. parts for legs and related terms o- ~ - Table 6 Seven basic zones in the partonomy of body

Since the INA are classificatory in their functions, we can specify the class of the possessed nouns. Thus INA 2 means noun-class-2 and INA 4 designates noun-class 4, so on and so forth in the grammar.

5.4 Parallels between body parts and kinship terms

There are only four out of the seven INA markings that are used to represent the kin terms. The four most commonly used are ut-, r, a-, and ara-. Table 6 summarizes the parallel relationship that exists between the body part terminology and the kin terminology.

Class Inalienability Body parts Kin terms prefix 1 a- mouth cavity parents 2 εr- major body parts spouse 4 ut- extensions of body parts child Or body products

6 ara- + GEN 2 nodular structure younger/older siblings Table 7 Parallel between body parts and kinship terms

18 5.5 Twin levels The possessives function at twin levels in PGA, namely the primary and the secondary. Those with a primary level of function are used with reference to the ‘self’, which denotes the major body divisions/areas and the main kinship terms that we just considered. Those with a secondary level of possession are used for denoting those body parts that are derived from the major parts and those kinship terms that are descriptive, e.g. the ones used for siblings (the fourth one in Table 6). The secondary divisions are added to the basic ones to represent the extension of the body parts discussed above. For example, the language uses double markings to refer to words for ‘eyelashes’, ‘tears’ etc. The kin terms, perceived as of secondary nature, are similarly marked.

5.5.1 Formation of secondary possession

In addition to simple prefixation of seven distinct INA markers to the possessed nouns the language offers two more strategies to derive inalienable possessive constructions: (1) compounding or juxtaposition of two nouns and, (2) syntactic derivation where a combination of more than two devices is used. Examples of compounding are given below:

5.5.1.1 Juxtaposition/Compounds In the examples given below the second noun, i.e. the head noun, designates a generic entity while the first noun which is a dependent noun, indicates the type or class to which the designated entity belongs. Great Andamanese Literal translation (12) cokbi tomu turtle meat

(13) cokbi mulu t urtle egg

(14) mc mulu hen egg

(15) kidr coconut tree

(16) kidr ino coconut water

(17) ko tei tree blood (gum)

19 (18) ra thire pig children (piglets)

5.5.1.2 Complex Structures The body part terms which are not basic but augmentable in nature are derived by several morphological processes.

(i) Lexical Compounding with INA

(19 ) [=εr=[tap bec]] chin hair ‘My beard.’

(20) [=εr=[juku bec]] above upper lip-hair ‘My moustache.’

(21) [=ara= [karap ѐ]] waist bone ‘My waist bone.’

(ii) Double marking Different parts and sub parts of the eye are not simply juxtaposed to the term for ‘eye’. Instead, the nouns in this category, which are subordinate to the ‘parta’ eye, obligatorily need the use of the suffix - tu, which literally means, ‘born of’ and has grammaticalized as the possessive marker in these constructions.

(22) =εr= ulu -tuino

1SG -INA 2-eye –born-water

‘My tears.’

(23) = εr = ulu -tubec

1SG -INA 2-eye- born - hair

’ My eyelashes.’

(24) lico-εr= ulu - tubk

Lico-INA 2- eye-born-behind

’ Licho’s eyelids.’

(iii) Adverbial modification

Symmetric kinship relationships, such as those that represent sibling relations, are derived by using an adverbial phrase ut-toa-thu ‘born before’ or ara-sulu-thu ‘born after’

20 where thu- ‘be born’ describes the temporal relationship between the dependent noun, i.e. the possessor and the head noun, i.e. the possessed.

(25) =ut = toa tu kaa

1SG =INA 4-before-born girl

‘My elder sister.’ (Literally: ‘Before me born girl.’)

(26) =ara =sulu - tu a

1SG =INA 6 –after- born boy

‘My younger brother.’ (Literally: ‘After me born boy.’) The secondary possession indicates that the nominal body part term on the left is the head and is also the possessor of the second body part term which is an extension of the former. This is true of the compound formations that were cited above and also by the following examples.

Double marking possessives can also be derived by employing two different INA markers in the same NP, without using adverbial phrase and the grammaticalized -thu

‘born of’. Consider the following where POSSESSIVE markings are progressively decided by the head noun. (27) =er =piletara =po

2SG =INA 2-teeth-INA 6-cavity ‘Your dental cavity.’

(28) =er =ulutut =ltmo

1SG =INA 2-eye-INA 4-white

'The white of my eye (sclera).’

(29) =er =jilitot =bc

1SG =INA 2- bone above eye-INA 4-hair

‘My eyebrows.’

(30) =erpiletara =tarale

21 1SG =INA 2-teeth-INA 6-on ( DEXIES OF CONTACT )

'My gums.'

(31) =er =kotara =po

1SG =INA 2-nose-INA 6-cavities

‘My nostrils.’

(32) =o=mtut =juku

1SG =INA 7-leg-INA 4-end

‘My toes.’ To summarize, there are twin levels of possession functioning in PGA, the primary and the secondary. The primary level is used with reference to the self, which denotes the divisions/areas of the body and the main kinship terms. The secondary level of possession is used for denoting those body parts that are derived from the major seven divisions/areas and those kinship terms that are relational in nature, e.g. the ones used for siblings. Affixation, juxtaposition/compounding and syntactic derivation are three processes that are employed in relating the possessor and the possessed nominals. These different processes in combination with the markers discussed above offer as many as eleven different forms of POSSESSIVES selected on the basis of ethno-semantic divisions in the language. It is a unique language where both the possessor and the possessed decide the form of the possessive affix. Another unique feature is that while alienable possessive affix, i.e. the GENITIVE is suffixed to the possessor the inalienable possessive affix, i.e. the INA is prefixed to the possessed in the form of proclitic. These are given in Table 8.

Number Word formation POSSESSION Semantic Hierarchical type and MORPHEMIC category & level STRUCTURE Noun class 1 Affixation a- ~ ta- Inalienable Primary 1 2 Affixation ϯr- ~ tϯr- Inalienable Primary 2 3 Affixation o- ~ to - Inalienable Primary 3

22 4 Affixation ut- ~ ot- ~ Inalienable Primary tut- ~ tot- 4

5 Affixation e- ~ te- Inalienable Primary 5 6 Affixation ara- ~ tara- Inalienable Primary 6

7 Affixation o- ~ - ~ Inalienable Primary to- ~ t - 7 8 Affixation -ico ~ -io Alienable Primary 9 Compounding Possessor Noun + Inalienable Secondary Possessed Noun 10 Compounding INA + N-N Inalienable Secondary with INA 11 Complex INA -Noun/Adv + GEN - Inalienable Secondary Double marking Noun Table 8 Varieties of possession in Great Andamanese

It is worth noting that out of eleven INA markers only one is used for alienable possession. The rest are seen as variations of inalienable possessions.

6. Other nouns

The INA markings are also used with nouns designating time, direction and depth, resulting in various deictic distinctions . Some examples are in order.

Levels of the sea (33) iro ’sea’ (34) iro-tr=ik ui/lik u 'Deep sea.' (35) iro-tr=kra 'Shallow sea.'

Phases of the sun and the moon (36) iu -tara=bat ‘dusk’ (37) iu-tara=cѐl ‘sunlight’ Location

(38) ɲo-ta=cѐko ‘Side of a house.’ (39) h-a=cѐko ‘My side.’ (around the armpit)

23 7. Alienable possession

There is only one genitive suffix attached to the pronominal root/noun in the language to denote alienable possession. Predictably, this is used only with the animate possessors and has two allomorphs; -ico ~ -io . Most of the typical alienable nouns designating ‘land’, ‘jungle’, ‘upper garments’, ‘lower garments’, ‘dog’, ‘friend’, ‘God’, as well as some kinship terms such as ‘son’, and ‘daughter’ are marked by alienable GEN -ico or -iЀo. Thus:

(40) uiockbi

3SG .DIST .VIS -GEN -turtle ‘His turtle.’

(41) n =ioko 3PL -GEN -bow ‘Their bow.’

(42) =icoboa 1SG -GEN -land ‘My land.’

(43) =icoimiku 1SG - GEN -forest ‘My forest.’

(44) =icoa cao 1SG -GEN -ARG dog ‘My dog.’ (45) =icojo 1SG - GEN -song ‘My song.’

(46) buli -icoo Buli-GEN -dress ‘Buli’s dress’ 8. Verbs The Great Andamanese grammaticalize the primary INA and use them in two ways in a sentence. (1) Verbs of the type ‘exit’, ‘leave’, ‘come’ and ‘go’ as well as many that are

24 experiential in nature dictate the primary INA marker of class 4, viz. -ut ~ -ot suffixed to the agent nominal or pronominal clitic. Thus, h=ut-cone-bom ‘I am going’ or thire-ut-

he e-bom ‘the child is hungry’. (2) A large number of Great Andamanese verbs are necessarily prefixed by an object clitic, the phonetic shape of these clitics is decided by the nature of the verb and the associated object seen in the context of the partonomy of the body. The first two of the following examples have INA 5 /e-/, signifying something internally possessed. In the last three examples, the different INA markings used with the verb ‘aim’ distinguish between the various ways of aiming at an object in a hunter- gatherer society.

(47) e=ro Ѐe ‘to love’ ( INA 5)

(48) e= Ѐuye ‘to roast’ ( INA 5)

(49) ara= Ѐuye ‘to cook by putting a pot on the fire’ ( INA 6)

(50) ut=tε֊ ‘to smell’ ( INA 4)

(51) ut= Ѐile ‘to aim from above’ ( INA 4)

(52) ek= Ѐile ‘to aim at’ (resultative) ( OBJECT CLITIC )

(53) e=Ѐile ‘to aim to pierce’ ( INA 5)

The verb ‘to see’ which is marked by internal body parts affix e- ~ - changes the meaning if prefixed by a relational body part affix r-.

r=ole =ole

INA 2-see INA 5-see ‘Call someone by gesture’ ‘see’

Verbs of intransitive nature can be prefixed with one of the INA MARKERS designating various psychological predicates, experience, and state. a=jet =colol e=bi e

INA 4-vomit like INA 5-roll INA 5-think/remember ‘feel nauseated or uneasy’ ‘roll down’ ‘think’

25 r=jome ut=jome

INA 2-scare INA 4- scare- PST ‘be afraid’ ‘get startled'

Thus, what strikes us most from the examples given above is that the basic division in verbs is not between +/- transitive but between +/- dependency. Verbs are either marked with INA or are unmarked. The marked ones can take any one of the seven INA MARKERS and/or an object clitic. Great Andamanese grammar helps us to extend the notion of ‘dependency’ to areas beyond nominals.

8.1 Verb Classes

Verbs belong to different classes represented by a distinct consonant 7 that precedes the mood or aspect and tense [MAT for short] categories. Let us term this consonant, a ‘thematic consonant’ because it determines the class of the verb. A verb in its bare form neither shows the class nor the MAT categories. In fact, unless a verb is inflected for MAT it is not possible to know what class it belongs to. The thematic consonant can never terminate the verbal form. It is always followed by a vowel, which indicates the aspect or mood. An optional consonant at the ultimate position indicates non-past tense. The absence of the terminal consonant signifies past tense. We could identify only the following six consonant classes of verbs.

–b or –l or –k or –r or –pܒ or, –m

The -k class is very common, and it appears that most of the verbs belong to this class. The verb schema could be as given below. No other Indian language has even a slight resemblance to such verb structures .

(S 3) Verb root + Cons Class + Vowel + Consonant/zero [Aspect/Mood] [Tense]

Thus: (54) mfe-k-o-m jump-CL -IND -NON PAST

7 Manoharan (1989) reports seventeen distinct suffixes, each a CVC type.

26 ‘(He) jumps.’

(55) ie-k-e give-CL -IMPERATIVE ‘Give it.’

(56) ek-khiu-k-o-ø OBJ .CLT -pour- CL -IND -PAST ‘(He) poured it once.’

8.2 Durative versus non-durative

The distinction between –b class and –k class verbs is not very clear. However, a speculation can be made that verbs which are inherently non-durative are marked by –k and those which are inherently durative, i.e. have the potentiality of being continued over a period of time are marked by –b. As the language is mixed and several verbs have been incorporated from various varieties of Great Andamanese, the current speakers use –b or –k markers indiscriminately. There is also a recent trend of dropping the consonant class altogether. Thus, iji-k-om ‘he eats’ can be rendered as iji-om in the present speech of the Great Andamanese.

It appears that –l class verbs have directional meaning. Thus ra u-l-om ‘he kicks’ or itbo-l- ‘he searched’. The –l marker is used even for death, as the person leaves this earth, as in εm-pi-l-o ‘s/he died’. The sentence uolopu ‘I did not walk’ has an -l class verb.

9. Inalienability and its representation on modifiers The case for adjectives and adverbs is similar to the case of verbs. A kind of semantic transparency, drawn on the basis of the original classificatory meaning assigned to the body part terms by the INA , can be seen between these markers and the host adjectives to a large extent. This, at times, leads to a certain amount of possibility of determining the choice of the adjective with a particular marker. For instance, we observe that the INA marker i- ~ e- is attached to those terms for body parts which are inside the body, e.g. ‘blood’, ‘intestines’, etc. and the same marker is attached to adjectives defining internal

27 human propensity such as in e-liu ‘brave’; e-cay ‘bad’; e-irim ‘black’ or ‘dark’;

bop o ‘stupid’. It may also signify internal quality of an inanimate object such as in ekok ʰela ‘blunt’; i-boe ‘boiled’; e-mello ‘thick’ and i-pu ‘fully ripe’.

9.1 Attributive Noun Phrases Attributive noun phrases are derived by using one of the INA markers reserved for inalienable possessions. At times double markers are also used as discussed earlier under secondary possession.

(57) hi-tѐt=bec-ta =pho ֊ ‘opening in a forest with little undergrowth’ (58) ino-tεr=pho ֊-ta =ino ‘water of a well’

Modifiers of verbs, viz. adverbs can be prefixed by these INA markings designating various deictic meanings as well as manner of an action. In this function these markers are highly grammaticalized.

10. Process of grammaticalization

It is clear by now that INA markers, each with a specific meaning, are grammaticalized in the language and co-occur with every grammatical category of content words classifying and modifying them. It is not easy to establish a one-to-one correspondence with these

INA markers and the INA markers used for body part terms but the native speakers of the language have no problem in assigning an appropriate one in the case of new adjectives, new verbs and new nouns. Although it is very difficult to distinguish one specific kind of meaning from the other while analyzing each of the seven divisions represented in verbs, adjectives, adverbs and nouns, we can represent the process of grammaticalization for each category considered so far. Refer to Table 9.

28 Classes Partonomy INA VERBS ADJECTIVES ADVERBS of human MARKERS body

1 mouth and a- mouth-related mouth-related deictic meaning its semantic activity, origin attributive of front or back, extension quality of a anteriority of an person, ‘dead’ action 2 major ϯr- activity in attribute of deictic meaning external which the size, external of adjacency, body parts front part of beauty uncontrollable the body is actions/emotions involved. 3 extreme o- hand-related attributes ‘hurriedly’ ends of the activity, related to body like toe action to do limbs, ‘alive’ and with fingernails extremities of body 4 bodily ut- directional, attributive emerging out of products and away from the quality of an something, deictic part-whole ego, X after a part meaning of relationship experiential is taken out of ‘towards X’ it 5 organs e-, - internalized inherent deictic meaning inside the action, when attribute of X of ‘in the middle body the effect of of X’, ‘slowly’ an action can be seen on the object, or experienced 6 parts ara- action that attribute of deictic meaning designating involves side size, ‘time’ of something in round shape or middle and belly- contact with portion of the related body 7 parts for leg o- ~ - action which external temporal deixis and related more often attribute of an relating to ‘sun terms than not, X rise’ or directional results in deixis roundish object or in a definite result

Table 9 Partonomy of human body and grammaticalization process

29 11. Ambivalence of grammatical categories Great Andamanese offers ambivalence of several grammatical categories. Thus, a phonological word can serve as a noun, a modifier or a verb, depending upon the word order or morphological terminations. For instance the word nl ‘good’ is realized as a verb if it is affixed with tense and mood markers as in nl-o-m ‘is doing good/well’, or as an adjective in o nl ‘good house’. Similarly, the word for ‘house’ o is rendered as a verb to mean ‘live’ in o-il ‘having lived’ or o-m ‘he lives’. The word kimil ‘friend’ can be used as an adverb or an adjective or as a verb depending upon the appropriate context.

12. Conclusion The language offers many surprises, especially in the realm of possessive constructions. The complexities in verb class and possession suggest that the language is very old and has evolved over thousands of years. Considering a large inventory of phonemes and variant forms it offers supporting evidence of being one of the oldest languages of the world (Refer to the latest discovery of computing antiquity by Atkinson 2011). If Andamanese are the remnants of the first migration out of Africa and have lived in isolation all throughout the history as maintained and proved by the geneticists (Kashyap et al 2003, Thangaraj et al 2005, 2006) then it will not be surprising to discover a few traits which are retained beyond the 8000 years-mark observed by historical linguists (Nichols 1992). The perception of the world through the body division and then grammaticalizing it through every grammatical category appears to be an archaic trait which also diffuses the strict dichotomy of nouns and verbs.

It can be argued that in the case of Great Andamanese, inalienable possession of anatomical terms is the basic semantic relation that gets symbolized in various proclitics. The possessor, in this context, is the human being---not the body. These proclitics attach to different words of various grammatical categories and render appropriate meaning. Since the aboriginal populations of the Andaman Islands have remained in isolation for a much long period than any known ancient population of the world

30 (Kashyap et al 2003), their language has retained some very ancient and unusual linguistic structures not shared by any other language known to this author.

*Acknowledgments I am grateful to the Great Andamanese speakers, past and present, who willingly contributed to the genesis of the article. I am also very thankful to my team members of the Pilot survey (2001-2002) Shailendra Mohan and Pramod Kumar as well as to the members of the VOGA team especially, Bidisha Som, Narayan Chaudhury, Abhishek Avatans and Mayank Jain.

I am grateful to Andrej Malchukov, Max Planck Institute of Evolutionary anthropology, Leipzig, Germany, who spent long hours discussing sections on grammaticalization. I benefited immensely by his suggestions. Parts of this work were presented in January 2011 at the James Cook University, Cairns, Australia. I am very thankful to Alexandra Aikhenvald and R.M.W. Dixon for their feedback and suggestions.

References Abbi, Anvita. 2003 Vanishing Voices of the Languages of the Andaman Islands, June 13, 2003. Paper presented at the Max Planck Institute, Leipzig. Posted in 2004 ‘The Great Experience: A Linguistic Fieldtrip to the Andaman Islands, 2001-2002’ A. Abbi www.andaman.org 2006 Endangered languages of the Andaman Islands. Munchen, Germany. 2009 Is Great Andamanese genealogically and typologically distinct from Onge and Jarawa? Language Sciences 31:6. 791-812.

2010 Semantics of inalienability and grammaticalization of body part terms in Great Andamanese . Forthcoming. 2011 A Dictionary of the Great Andamanese language. English-Andamanese- Hindi with sound files. Delhi. Ratna Sagar (in Press).

31 Atkinson, Quentin, D. 2011 Language Expansion from Africa Phonemic Diversity Supports a Serial Founder Effect Model of Language Expansion from Africa. Science . Vol.332, 15 April 2011. DOI: 10.1126/science.1199295 Blevins, Juliette. 2007 A long lost sister of Proto-Austronesian? Proto-Ongan, mother of Jarawa and Onge of the Andaman Islands. Oceanic Linguistics 46 (1): 155–198 . Endicott, Phillip, Thomas, M., Gilbert, P., Stringer, Ch., Lalueza-Fox, C., Willerslev, E., Hansen, A.J., Cooper, A. 2003 The genetic origins of the Andaman Islanders. American Journal of Human Genetics 72, Report no. 178. Hagelberg, Erika, Singh, Lalji, Thangaraj, K., Reddy, A.G., Rao, V.R., Sehgal, S.C., Underhill, P.A., Pierson, M., Frame, I.G., 2003 Genetic affinities of the Andaman Islanders. A vanishing human population. Current Biology 13, 86–93.

Kashyap, V.K., Sitalaximi, T., Sarkar, B.N., Trivedi, T.

2003 Molecular relatedness of the aboriginal groups of Andaman and Nicobar Islands with similar ethnic populations. International Journal of Human Genetics 3(1): 5-11. Manoharan, S. 1989 A Descriptive and Comparative Study of the Andamanese Language. Calcutta. Anthropological Survey of India. Nichols, Johanna. 1992 Linguistic Diversity in Space and Time . Chicago. University of Chicago Press.

Siegel, J. 1985 Koinès and Koinèization. Language and Society . 14, 357-378.

32 Thangaraj, K., Singh, L., Reddy, A.G., Rao, V.R., Sehgal, S.C., Underhill, P.A., Pierson, M., Frame, I.G., Hagelberg, E. 2003 Genetic affinities of the Andaman Islanders. A vanishing human population. Current Biology 13, 86–93.

Thangaraj, K. Gyaneshwer Chaubey, Toomas Kivisild, Alla G. Reddy, Vijay Kumar Singh, Avinash A. Rasalkar, Lalji Singh. 2005 Reconstructing the origin of Andaman Islanders. Science 308, 996.

Thangaraj, K. Gyaneshwer Chaubey, Toomas Kivisild, Alla G. Reddy, Vijay Kumar Singh, Avinash A. Rasalkar, Lalji Singh.

2006 Response to comment on ‘‘Reconstructing the Origin of Andaman Islanders”. Science 311, 470b.

Websites consulted • www.andaman.org

33